The full receipts archive, citable.
Every receipt we've published, exported as JSON and CSV at stable URLs. MIT-licensed. Updated within an hour of the next published receipt. Built so journalists, researchers, procurement teams, and AI agents can cite one source for “what changed at AI vendors and when.”
DOWNLOADS 3 endpoints · CORS-open · ETag & Last-Modified
Every receipt with vendor name, severity, delta, sources, AI Jury aggregate. The canonical shape - same schema served at /api/v1/receipts.
{ "meta": { "version": "2026-06-22", ... }, "receipts": [ { "id": "cursor-2025-06-credit-rebill", "vendor": "Cursor", "severity": "critical", "nerfScore": 91 ...
Same data, flat. Drops into Sheets / Excel / R / Stata without preprocessing. Source citations are pipe-delimited in the sources column.
id,date,vendor,severity,nerfScore,kind,... cursor-2025-06-credit-rebill,2025-06-14,Cursor,critical,91,... replit-2025-09-effort-pricing,2025-09-02,Replit,critical,88,... copilot-2025-12-modernized,2025-12-08,GitHub Copilot,major,67,... ...
Per-vendor Nerf Index, receipt counts by severity, last-event timestamps. Pre-aggregated so you don't have to roll up the full corpus to display a leaderboard.
{ "vendors": [ { "slug": "cursor", "nerfIndex": 91, "band": "predator", "counts": { "critical": 14, ... }
QUICK START curl + jq · plain HTTP, no auth required
# 1 - Full corpus + metadata header $ curl gotnerfed.com/data/receipts.json | jq '.meta' # 2 - CSV mirror - drop into Sheets / Excel / R $ curl -o gotnerfed-receipts.csv gotnerfed.com/data/receipts.csv # 3 - Per-vendor index, sorted by Nerf Index descending $ curl gotnerfed.com/data/vendors.json | jq '[.vendors[] | {slug, nerfIndex}] | sort_by(-.nerfIndex)' # 4 - Only critical receipts in the last 90 days $ curl gotnerfed.com/data/receipts.json \ | jq '[.receipts[] | select(.severity == "critical" and .date > "2026-05-30")]'
CITATION suggested wording · stable across versions
GotNerfed receipts dataset (https://gotnerfed.com/data). Updated 2026-06-22. MIT license.LICENSE & TERMS
The dataset is MIT-licensed - we just ask for attribution back to gotnerfed.com. We do not warrant accuracy of individual receipts beyond what's in the methodology. Every receipt cites at least one primary source; AI Jury verdicts are aggregated from independent LLM passes and disagreements are visible per-receipt.
For corrections, email corrections@gotnerfed.com. Schema is versioned via the meta.schemaVersion field; breaking changes get a major version bump and a 30-day deprecation window on the prior shape.
SCHEMA 12 fields · full descriptions in meta.schemaFields
cursor-2025-06-credit-rebill. Permalink id; never reassigned.band ∈ saint | clean | watch | shrinking | predator.billing-model-shift · rate-limit-cut · plan-tier · model-swap · tier-removed · feature-gated · price-increase · idle-burninfo · minor · major · critical.lib/score.ts pipeline.{ count, harmRulings, avgScore }. 3-LLM AI Jury aggregate. null until the jury runs (typically within 6h of publish).